My guess is that there were several bug fixes in the AWT native interface stuff in version 1.4.2. Even though the AWT native interface stuff has been in Java since 1.3 it still seems like it is relatively new, i.e. has a few bugs. Or perhaps it is an incompatibility between the shared libraries on your system and the shared libraries that the JDK was built with. I know that I had issues with a program that did very intensive Java2D, it would crash the JVM. When I upgraded from Redhat 7 to Redhat 8 (or it might of been 8 to 9), then the problem went away–more than likely a native library upgrade fixed the problem.
The section of code that it is dying on is the native glXMakeCurrent is returning false. More than likely one of those parameters is incorrect. The first thing I would check is the FIXME regarding 64-bit systems. If that is not the issue, then I would look into where the values for those parameters are coming from and ensuring that they are retrieved in the proper way. I had the exact same (or similar) error on a multiheaded linux system when I was trying to show JOGL on a screen other than the primary. It turned out that JOGL defaulted to screen 0 which was causing the an incorrect value in the call to glXMakeCurrent. The solution to this problem was to get the default graphics device from the graphics environment and figure out what screen that was. (If you had an application that needed to be multiheaded, you could set the graphics configuration in the GLCapailities to select which screen the window should go on.) I’ve submitted my fix to Sun, but we haven’t heard anything back yet. Your problem may be related to this.